I build production AI agents for the enterprise — and a system around how I build them.
At ADP I work on ADP Assist, the company's flagship AI assistant, building
semi-autonomous looping agents that call MCP tools across thousands of API endpoints to make
multi-step decisions in real time.
Ten years in conversational AI, six of them at enterprise scale. I've carried the same product
across a full architectural generation — from an IBM Watson Assistant platform to an
event-driven TypeScript stack with LangGraph, and on to agent frameworks including AWS Strands.
Before the technical career I spent a decade in operations, ecommerce and C-suite strategy,
which is why I get pulled into ambiguous problems where nobody has written the requirements yet.
Selected Achievements
Led the migration of all 65 IBM Watson Assistant agents to a TypeScript event-driven
architecture at ADP, preserving functionality across Benefits, Payroll, Time Off, Total Absence
Management, Hiring, Personal Profile and Wisely Pay.
Built and maintained conversational AI serving tens of millions of users monthly
on ADP's HCM platform.
Delivered $7 per call in cost savings across millions of calls at Aetna / CVS Health,
replacing live-agent handling on their voice IVR system.
Designed and shipped ADP's employee profile Strands agent — a semi-autonomous looping
LLM replacing a simple request/response flow, making multi-step permission, intent and
retrieval decisions per interaction via MCP tools.
Roughly doubled personal delivery capacity through a spec-driven, wiki-grounded
workflow with parallel agent sessions — an approach project managers began seeking him out
to learn.
Secured an exclusive IBM license for proprietary far-field video optics IP, extending
facial recognition range in IBM Intelligent Video Analytics.
Shipped an independent product through Google Workspace Marketplace review —
OAuth scopes, verification, privacy policy and security assessment, end to end.
Experience
Senior Lead Application Developer, AI Assistant
ADP — RemoteApr 2025 – Present
Builds AWS Strands agents for ADP Assist — looping LLM architectures calling MCP tools
across thousands of API endpoints.
Designed the employee profile agent: permission resolution, intent disambiguation with
clarification prompts, MCP tool retrieval, and an edit path — each loop iteration driven by
the model rather than scripted.
Contributor to ADP's internal Claude Code adoption initiative — parallel agent sessions in
VS Code connected to Jira and Bitbucket for implementation, test authoring, ticket
documentation and CI troubleshooting.
Built a persistent, self-updating engineering wiki plus a custom ingestion skill, giving
agents compounding context across actively-developed brownfield codebases.
Produces single-page, accessibility-aware HTML status reports for multi-stakeholder projects.
Senior Application Developer, Chatbot
ADP — Remote (contract via Advanced Tech Placement, converted to FTE)May 2022 – Apr 2025
Led the full migration of 65 IBM Watson Assistant agents into TypeScript within an
event-driven architecture, modernizing the platform without losing functionality.
Built and maintained enterprise conversational AI serving tens of millions of users monthly
across payroll, benefits, absence management and time domains.
Prompt engineering and accuracy auditing on ADP's RAG system (AWS + Anthropic Claude):
audited responses intent-by-intent, refined prompts iteratively, validated with BLEU and
ROUGE scoring in Databricks. Prompt engineering and audit role — did not
build the embedding or retrieval pipeline.
Contributor to ADP's "Transformative Products" recognition in the 2025 BIG Innovation Awards.
A.I. Modeler / Senior Application Developer, Conversational AI
Aetna, a CVS Health Company — Remote (contract)2021 – 2022
Delivered $7 per call in savings on millions of calls previously handled by live agents on
the voice IVR system.
Built Watson dialog logic in IBM's proprietary language, connecting backend microservices to
return personalized, entitlement-aware responses.
Modeled and refined NLU intent classifiers and entity extractors; authored playbooks on
Ground Truth Creation methodology.
Managed the Git repository and CI/CD pipeline documentation; served as primary on-call
production support.
Conversational AI UX Designer
Lincoln Financial Group — Remote (contract)Jan 2020 – Feb 2021
Designed voice and chatbot experiences on Google Dialogflow for the life insurance division,
targeting reduced live-agent call volume.
Analyzed call transcripts and recordings to establish what callers actually contacted about,
driving intent design and handoff decisions.
Annotation and labelling for intent classification and entity extraction.
Analyzed conversational performance in Tableau; co-authored enterprise bot framework
guidelines covering governance and security.
NLP Specialist — Independent AI Consulting (Artificial Savant)
Self-employed · Registered IBM Business PartnerApr 2016 – Dec 2020
Promoted and demonstrated chatbots and video analytics to businesses of every size,
operating under the Artificial Savant brand.
Secured an exclusive license with IBM to bundle proprietary video optics IP with IBM
Intelligent Video Analytics.
Developed a long-distance facial recognition program achieving roughly 400% greater depth
of field in surveillance video.
Designed a conversational loan application supporting dynamic interviews across users with
differing financial histories, and an AI voice agent that interviews job applicants.
Earlier: Business Operations & Strategy
ABC Water · Digitel · Humes Ford · M&M Design & Building · Benefit Commerce Group2004 – 2019
Chief Strategy Officer / UX at ABC Water (2016–2019) — 300% online revenue growth,
$1.5M year-over-year variance.
Director of Ecommerce & UX at Digitel (2014–2016) — 100% year-over-year sales growth.
Finance Manager, VP of Operations, and employer benefits strategy roles with P&L
ownership and team leadership.
Skills
Conversational AI
IBM Watson Assistant, Google Dialogflow, TypeScript dialog flows, intent design, NLU modeling,
entity extraction, ground truth creation, VUI design
Premium Tracker — Google Workspace Marketplace add-on for options traders. Taken
through full marketplace approval.
Marketplace listing
SEC EDGAR Analysis Platform — XBRL parsing from SEC filings into a nine-category
value-investing scoring model. GitHub
Full Semester — AI edtech tool extracting dates from unstructured syllabi into a
colour-coded Google Calendar. Founded; did not reach sustainable traction.
Painted UI — ongoing research into reproducing generative-UI motion at near-zero
marginal cost per user. GitHub
Education
B.S. Business Administration — Management Information Systems